What it is.
The brief carries about 70% of the draft quality here, which is why the guide opens on brief sections three, five and nine, the ones teams skip most.
The six-stage chain runs scaffold, elaborate, pressure-test, cite, compress, polish. Drafts from the chain pass editor fact-check on first read about 85% of the time; single-prompt drafts on the same briefs pass closer to 55%.
The honest part is the editor ratio. Roughly 90% of sentences get rewritten before publish, while about 70% of the information survives.
What you get.
- Which brief sections carry draft quality, and which get skipped
- The voice rules file, and why voice failures are patterned
- A six-stage prompt chain that holds a 3,000-word piece together
- First-read fact-check rates of about 85% for the chain against roughly 55% for one prompt
- The editor ratio: about 90% of sentences rewritten, roughly 70% of information kept
- The red flags editors never let through
Read it at the source.
Read the guide
Why not a single prompt?
How much editing should I budget?
Can AI write publishable long-form content?
Where do AI drafts hallucinate?
Why use a prompt chain over one long prompt?
How much does the brief matter?